Reciprocal Relations Between Recovery and Work Engagement: The Moderating Role of Job Stressors

Abstract

In this paper, we examined the within-person relations between morning recovery level (i.e., feeling refreshed and replenished) and work engagement throughout the day, and between work engagement throughout the day and the subsequent recovery level at the end of the workday. We hypothesized that job stressors (situational constraints, job demands) moderate these relations. A diary study over 1 workweek with 2 measurement occasions per day (N = 111 persons) provided support for most of the hypotheses: Morning recovery level predicted work engagement, and work engagement predicted subsequent recovery level at the end of the workday after controlling for morning recovery level. As predicted, situational constraints attenuated these relations, but job demands did not. The results suggest that recovery translates into employee work engagement, and work engagement, in turn, prevents a loss in recovery level throughout the day, particularly when situational constraints are low. Situational constraints seem to interrupt the reciprocal processes between recovery level and work engagement.

Full text

Reciprocal Relations Between Recovery and Work Engagement:
The Moderating Role of Job Stressors
Sabine Sonnentag
University of Mannheim
Eva J. Mojza
University of Konstanz
Evangelia Demerouti
Eindhoven University of Technology
Arnold B. Bakker
Erasmus University Rotterdam
In this paper, we examined the within-person relations between morning recovery level (i.e., feeling
refreshed and replenished) and work engagement throughout the day, and between work engagement
throughout the day and the subsequent recovery level at the end of the workday. We hypothesized that
job stressors (situational constraints, job demands) moderate these relations. A diary study over 1
workweek with 2 measurement occasions per day (N!111 persons) provided support for most of the
hypotheses: Morning recovery level predicted work engagement, and work engagement predicted
subsequent recovery level at the end of the workday after controlling for morning recovery level. As
predicted, situational constraints attenuated these relations, but job demands did not. The results suggest
that recovery translates into employee work engagement, and work engagement, in turn, prevents a loss
in recovery level throughout the day, particularly when situational constraints are low. Situational
constraints seem to interrupt the reciprocal processes between recovery level and work engagement.
Keywords: diary, job demands, recovery, situational constraints, work engagement
Work engagement, defined as a positive and fulfilling work-
related state of mind (Schaufeli & Bakker, 2004), is a pleasurable
experience for many workers. It goes along with feelings of
energy, dedication, and absorption in one’s work (Bakker,
Schaufeli, Leiter, & Taris, 2008; Schaufeli, Salanova, Gonza´les-
Roma´, & Bakker, 2002) and is associated with good mental health
and an increase in job resources over time (Schaufeli, Bakker, &
van Rhenen, 2009; Schaufeli, Taris, & van Rhenen, 2008). Work
engagement benefits the organization by stimulating task and
contextual performance (Halbesleben, Harvey, & Bolino, 2009;
Salanova, Agut, & Peiro´, 2005; Sonnentag, 2003). Moreover, it is
negatively related to withdrawal behavior (Newman, Joseph, &
Hulin, 2010; Schaufeli et al., 2009).
Although a person’s general level of work engagement is fairly
stable over time (e.g., Mauno, Kinnunen, & Ruokolainen, 2007), a
person’s day-specific level of work engagement fluctuates sub-
stantially around a person’s average level of work engagement
(Sonnentag, Dormann, & Demerouti, 2010). There are days when
a person who is highly engaged, on average, experiences a low
level of work engagement—and also days when a generally not-
so-engaged person is highly engaged. These day-specific varia-
tions of work engagement within persons are not arbitrary fluctu-
ations but can be explained by day-specific experiences and events
(Bledow, Schmitt, Frese, & Ku¨hnel, 2011; Ku¨hnel, Sonnentag, &
Bledow, 2012), and they predict systematic variations in outcomes,
such as proactive behavior and financial returns (Sonnentag, 2003;
Xanthopoulou, Bakker, Demerouti, & Schaufeli, 2009b).
Day-specific states, such as a person’s morning recovery level
(i.e., the experience of being refreshed and replenished; cf. Bin-
newies, Sonnentag, & Mojza, 2009), predict work engagement
throughout the day (Ku¨hnel et al., 2012). However, little is known
about the affective and energetic consequences of work engage-
ment. The question of whether work engagement is a refreshing
and replenishing experience in itself—as opposed to a process that
results in depletion— has been neglected so far. Our study ad-
dresses this gap in the literature. By examining both the relation
between morning recovery level and work engagement throughout
the day and the relation between work engagement and recovery
level at the end of the workday, our study investigates the recip-
rocal relation between recovery level and work engagement.
Such a reciprocal relation, however, may not be present on all
days. On days when employees face situational constraints (i.e.,
hindrance stressors that interfere with task completion), work
engagement may not benefit from a high morning recovery level
and may even become irrelevant for the recovery level at the end
of the workday. Thus, situational constraints may attenuate the
This article was published Online First April 30, 2012.
Sabine Sonnentag, Department of Psychology, University of Mannheim,
Mannheim, Germany; Eva J. Mojza, Department of Psychology, Univer-
sity of Konstanz, Konstanz, Germany; Evangelia Demerouti, Department
of Industrial Engineering and Innovation Sciences, Eindhoven University
of Technology, Eindhoven, the Netherlands; Arnold B. Bakker, Depart-
ment of Work and Organizational Psychology, Erasmus University Rot-
terdam, Rotterdam, the Netherlands.
This research was funded by a research grant (AFF 12/00) from the
University of Konstanz to Sabine Sonnentag. This grant is gratefully
acknowledged. We thank Franziska Bertram and Sonja Riefer for their help
during data collection.
Correspondence concerning this article should be addressed to Sabine
Sonnentag, Department of Psychology, School of Social Sciences, Univer-
sity of Mannheim, Schloss EO 245, D-68131 Mannheim, Germany.
E-mail: sonnentag@uni-mannheim.de
Journal of Applied Psychology © 2012 American Psychological Association
2012, Vol. 97, No. 4, 842–853 0021-9010/12/$12.00 DOI: 10.1037/a0028292
842

reciprocal relation between recovery level and work engagement.
However, high job demands, which are another type of job stres-
sor, may not have such a detrimental effect but may actually
enhance the association between morning recovery level and work
engagement. Our study examines situational constraints and job
demands as moderators in the relation between recovery level and
work engagement. We choose these stressors because they proto-
typically represent hindrance and challenge stressors as two core
dimensions of job stressors (Podsakoff, LePine, & LePine, 2007).
Taken together, our study has two aims. First, we examine the
dynamics between recovery and work engagement over the work-
day. Second, we examine job stressors as moderators in the rela-
tion between (a) morning recovery level and subsequent work
engagement and (b) work engagement during the workday and
recovery level at the end of the workday. Figure 1 shows our
conceptual model.
We seek to make three contributions to the literature. First, we
seek to extend previous day-level studies that have focused on
performance-related outcomes of work engagement (Xanthopou-
lou et al., 2009b). In order to gain a better understanding of the
potential consequences of work engagement, we look at recovery
level as an affective and energetic state that might result from work
engagement. As many employees face family responsibilities
when they come home from work, it is important to know if high
engagement at work results in a low recovery level after work,
which would compromise effective functioning in the family (Hal-
besleben et al., 2009). Second, our study looks at reciprocal
relations between recovery level and work engagement. It thereby
adds to the growing number of studies on gain cycles and spirals
associated with work engagement (Bakker & Bal, 2010; Salanova,
Llorens, & Schaufeli, 2011) by explicitly adopting a day-level
perspective. Third, by testing job stressors as moderators of the
relation between recovery level and work engagement, we exam-
ine when such reciprocal processes may break down. By demon-
strating that job stressors may stop reciprocial processes, our study
suggests one reason why the onset of gain cycles is hindered and
why—as a consequence—real gain spirals (i.e., an increase in
work engagement over time) are rarely found (Salanova,
Schaufeli, Xanthopoulou, & Bakker, 2010; Xanthopoulou, Bakker,
Demerouti, & Schaufeli, 2009a). We go beyond earlier day-level
research that has mainly examined job stressors as predictors of
strains (Rodell & Judge, 2009) and address the moderator effect as
an additional mechanism by which job stressors affect people at
work.
Core Concepts
Recovery Level
Employees’ daily lives can be described as cycles of work and
rest (Zijlstra & Cropley, 2006). During work periods, employees
exert effort that results in strain, while during rest periods (e.g.,
work breaks, free evenings) they recover from the previous strain
and their physiological and psychological systems return to a more
relaxed state. These cycles of work and rest imply that a person’s
recovery level (i.e., the person’s momentary state of feeling re-
freshed and replenished; cf. Binnewies et al., 2009) changes during
the course of the workday, with comparably high recovery levels
in the morning before work and low recovery levels at the end of
the workday. In addition to these within-day differences, an em-
ployee’s morning recovery level might fluctuate from day to day
(Sonnentag, 2003). There are days with high morning recovery
levels and other days with low morning recovery levels. Similarly,
the recovery level at the end of the workday should also fluctuate
from day to day. Conceptually, recovery level is related to strain
because both constructs reflect a person’s momentary state in the
cycle of work and rest (Zijlstra & Cropley, 2006). However,
recovery level and strain are distinct in that recovery level is the
result of preceding leisure-time experiences and sleep, whereas
strain is the result of preceding stress experiences. Empirically,
recovery level and strain indicators are negatively related, but the
correlations are relatively low, suggesting clearly distinct con-
structs (Sanz-Vergel, Demerouti, Moreno-Jime´nez, & Mayo,
2010).
Work Engagement
Schaufeli and Bakker (2004) described work engagement as “a
positive, fulfilling, work-related state of mind that is characterized
by vigor, dedication, and absorption” (p. 295). Vigor implies being
energetic and mentally resilient at work and being willing to invest
effort and to persist when difficulties arise. Dedication means
being enthusiastic and inspired at work and experiencing signifi-
cance, pride, and challenge. Absorption can be described as full
concentration at work and as the experience of being happily
engrossed in one’s work. Work engagement not only differs be-
tween persons but also fluctuates within persons from day to day
(Sonnentag, 2003; Xanthopoulou, Bakker, Heuven, Demerouti, &
Schaufeli, 2008). Whereas recovery level is a context-free state
Figure 1. Conceptual model. Numbers refer to the hypotheses.
843
RECOVERY AND WORK ENGAGEMENT

that exists regardless of whether one is at work or not, work
engagement is job related and refers specifically to the person’s
state while at work.
Job Stressors: Job Demands and Situational
Constraints
Job demands refer to job features that ask for sustained physical
or mental effort (Demerouti, Bakker, Nachreiner, & Schaufeli,
2001). Work overload and time pressure are typical examples of
job demands that are present in many jobs. Job demands are
associated with short-term strain symptoms, such as affective
distress, exhaustion, and fatigue (Ilies, Dimotakis, & De Pater,
2010; Zohar, Tzischinski, & Epstein, 2003), and longer-term
health impairments (de Lange, Taris, Kompier, Houtman, &
Bongers, 2003). However, high demands are not always detrimen-
tal for job performance (LePine, Podsakoff, & LePine, 2005) and
may even stimulate work engagement (Bakker, van Emmerik,
Geurts, & Demerouti, 2008).
Situational constraints are job features that make it difficult for
employees to translate their abilities and motivation into effective
job performance (Peters, O’Connor, Eulberg, & Watson, 1988).
Typical situational constraints include faulty equipment, missing
or inappropriate material and supplies, and missing or outdated
information (Peters & O’Connor, 1980). Situational constraints
can be seen as typical hindrance stressors that make effective
performance difficult or impossible (Podsakoff et al., 2007). Sit-
uational constraints have been shown to be positively related to
strain symptoms (Leitner & Resch, 2005; Spector & Jex, 1998)
and negatively related to job performance (Gilboa, Shirom, Fried,
& Cooper, 2008).
Development of Hypotheses
Morning Recovery Level as Predictor of Work
Engagement
Being recovered in the morning implies feeling refreshed and
having recuperated from previous strain experiences (Binnewies et
al., 2009). It means being in a cognitive, emotional, and physical
state that helps one to function well and to persist when difficulties
occur. When one is recovered, the fatigue level is low and no other
strain symptoms interfere with the work process. Such a state of
being recovered enables employees to become fully immersed in
their work and to fully concentrate on it. In line with a resource-
based view of work engagement and work behavior (Trougakos &
Hideg, 2009; Xanthopoulou et al., 2009b), being recovered is
associated with the availability of energetic and affective resources
that, in turn, facilitate work engagement. Thus, a high recovery
level should be associated with a high level of work engagement.
That is, if the employee arrives at work in the morning in a
well-recovered state, it will be more likely for the employee to be
energetic, enthusiastic, and persistent (i.e., to be highly engaged)
during the workday. Earlier research identified a person’s recovery
level in the morning as a predictor of work engagement experi-
enced during that day (Ku¨hnel et al., 2012; Sonnentag, 2003). In
our study, we want to replicate these findings and propose the
following as our first hypothesis:
Hypothesis 1: Day-specific recovery level in the morning is
positively related to day-specific work engagement during the
day.
Job Stressors as Moderators in the Relation Between
Morning Recovery Level and Work Engagement
We propose that the association between morning recovery level
and work engagement throughout the day is not uniform across all
days. Contingent on the level of job demands and situational
constraints present during the day, the association between recov-
ery level and work engagement will be stronger or weaker.
With respect to job demands, we propose that a high level of job
demands enhances the association between morning recovery level
and work engagement. Overall, job demands tend to be positively
related to work engagement (Crawford, LePine, & Rich, 2010).
Evidence from day-level studies supports the idea of a positive
association between job demands and work engagement (Bakker,
van Emmerik, et al., 2008; Ku¨hnel et al., 2012), suggesting
that—at least in the short run— high job demands and high work
engagement are not incompatible. High job demands imply that
one has a lot to do and is required to work intensely and to dedicate
much effort to one’s work. On days when job demands are high,
one can use the energy provided by recovery to become engaged
and absorbed in one’s work. In other words, when faced by high
demands, a high recovery level helps one to feel energetic and to
dedicate oneself to one’s work. On days when demands are low,
however, the need for effort investment is low. On such days, a
high recovery level cannot translate into work engagement because
there are no demands for which the energy available could be used.
As a consequence, a high recovery level will not be associated with
a high level of work engagement.
Hypothesis 2: Day-specific job demands moderate the rela-
tion between day-specific recovery level in the morning and
day-specific work engagement during the day. The relation
will be stronger when job demands are high than when job
demands are low.
We predict that situational constraints attenuate the relation
between morning recovery level and work engagement throughout
the day. In particular, when situational constraints are high, a high
recovery level in the morning should not translate into high work
engagement. There are at least two reasons for this assumed
moderator effect. First, situational constraints consume the energy
that has been provided by recovery. For example, when situational
constraints occur (e.g., when necessary information is missing),
energy has to be invested in order to overcome the constraints
(e.g., searching for information), leaving less energy that can be
invested into the task accomplishment process. Empirical evidence
from day-level research suggests that after having encountered
situational constraints at work, employees experience a reduction
in vigor (Sonnentag & Jelden, 2009). Thus, situational constraints
use up the energy provided by recovery; thereby, the vigor com-
ponent of work engagement is reduced. Second, situational con-
straints draw attention to these constraints and interrupt the work
process. Thus, when one faces situational constraints, it is more
difficult to become fully immersed in work processes because a
part of the attentional resources is directed to the constraint and not
844 SONNENTAG ET AL.

to the task. Thereby, situational constraints inhibit the experience
of absorption and dedication, making full work engagement un-
likely. We propose that situational constraints attenuate the rela-
tion between morning recovery level and work engagement. Feel-
ing refreshed and recovered in the morning does not result in high
work engagement when situational constraints are high.
Hypothesis 3: Day-specific situational constraints moderate
the relation between day-specific recovery level in the morn-
ing and day-specific work engagement during the day. The
relation will be weaker when situational constraints are high
than when situational constraints are low.
Work Engagement as Predictor of Recovery Level at
the End of the Workday
In addition, we propose that work engagement throughout the
workday is positively related to a person’s recovery level at the
end of the workday. Recovery level at the end of the workday
reflects the amount of resources that remain after having worked
the whole day. When one feels vigorous and is dedicated and
absorbed in one’s work, intrinsic work motivation is high (van
Beek, Taris, & Schaufeli, 2011). Intrinsic motivation implies that
the main motivational force is the activity itself, which is perceived
to be enjoyable and satisfying (Ryan & Deci, 2000). In a state of
high intrinsic motivation, it is easy to concentrate on the task and
to show goal-directed behavior (Shirom, 2011). Thus, one can
accomplish one’s work without having to mobilize additional,
compensatory effort or additional self-regulatory resources, which
would result in fatigue later on (Hockey, 1997). In other words,
when engaged at work, no— or only a few extra—resources must
be mobilized to get the work done. As a consequence, resources
will be maintained over the course of the workday, strain symp-
toms will not increase substantially, and the recovery level will
stay relatively high as opposed to states of low work engagement,
when strain accumulates over the course of the workday (Simbula,
2010). Day-level research on flow—an experience that is similar,
although not identical to work engagement—suggests a negative
association between flow and strain symptoms (Demerouti, Bak-
ker, Sonnentag, & Fullagar, 2012). Thus, one should still feel
relatively refreshed after a high-engagement workday compared to
a low-engagement workday.
Hypothesis 4: Day-specific work engagement during the day
is positively related to day-specific recovery level at the end
of the workday.
Situational Constraints as Moderators in the Relation
Between Work Engagement and Recovery Level
at the End of the Workday
The hypothesized positive association between work engage-
ment throughout the day and recovery level at the end of the
workday, however, will not occur on all days. We propose that it
will be attenuated by situational constraints but not by job de-
mands. Above, we have argued that it will be difficult to experi-
ence work engagement when facing situational constraints. How-
ever, when looking at a workday as a whole, there might be days
when both work engagement and situational constraints occur. For
example, one might start the day with a high level of engagement
and face situational constraints later (cf. Beal, Weiss, Barros, &
MacDermid, 2005). It might also be that in rare instances, work
engagement occurs despite the presence of situational constraints.
In such a situation, it is necessary to deliberately ignore the
constraints, if possible, in order to become absorbed in one’s work.
Thus, on days when situational constraints co-occur with high
work engagement or follow an episode of work engagement, the
benefits of work engagement for one’s recovery level will be
reduced. The low-strain situation enabled by a high level of work
engagement will be disrupted when confronted with situational
constraints that occur afterward. Moreover, attempts to ignore
these constraints will be effortful and will consume energy re-
sources. As a consequence, the low strain level cannot be pre-
served over the day and the recovery level will be reduced at the
end of the workday.
With respect to job demands, however, we do not expect a
moderator effect. Of course, job demands require that employees
exert effort, which might result in a reduced recovery level. De-
spite this potential main effect of high job demands, job demands
will not affect the hypothesized association between work engage-
ment and recovery level. The experience of vigor will not be
interrupted by high demands because the high energy level can be
used directly for addressing the demands. Similarly, high demands
will not interfere with dedication and absorption because the
demands and the engaging experience pull the employee in the
same direction that preserves energy resources. These processes
imply that an employee’s recovery level can still benefit from
work engagement when job demands are high. Thus, we do not
propose a moderator effect of job demands.
Hypothesis 5: Day-specific situational constraints moderate the
relation between day-specific work engagement during the
workday and recovery level at the end of the workday.
The relation will be weaker when situational constraints are high
than when situational constraints are low.
Method
Procedure and Participants
We recruited study participants by approaching organizations in
a variety of industries (services, production, administration, bank-
ing, insurance) by phone and requesting support for our study.
After managers expressed their respective organizations’ willing-
ness to participate in the study, we informed employees of these
organizations via e-mail about the project and invited them to sign
up individually for the study. We offered feedback about the study
results after completion of data collection as an incentive for
participation.
Participants were asked to complete a general survey and daily
surveys implemented on handheld computers. Participants were
instructed to complete the daily surveys in the morning before
leaving home for work (Tuesday through Friday) and in the
evening immediately after returning home from work (Monday
through Friday). On average, participants completed morning sur-
veys at 6:58 (SD !0.58 hr) and end-of-workday surveys at 5:25
(SD !2.02 hr). Participants filled in the general survey before
starting with the daily surveys. They completed the daily surveys
845
RECOVERY AND WORK ENGAGEMENT

during a single week, albeit the specific weeks differed across
participants.
The general survey was completed by 122 persons. Out of these
122 participants, 117 persons provided a total of 424 day-level
morning data sets (Tuesday through Friday) and a total of 528
end-of-workday data sets (Monday through Friday). Time stamps
recorded on the handheld computers indicated that 12 morning
surveys and 97 end-of-workday surveys were completed at wrong
times (e.g., morning surveys completed in the afternoon) and were
therefore excluded from the data sets, leaving 412 valid morning
data from 116 persons (on average 3.55 days per person) and 431
valid end-of-workday data from 114 persons (on average 3.78 days
per person). In the next step, we matched valid morning data with
valid end-of-workday data of the same day, thereby discarding 80
end-of-workday data assessed on Monday (because no morning
data were assessed on Monday) and leaving 351 valid end-of-
workday data from 112 persons. Matching valid morning and
end-of-workday data resulted in our final data set of 325 days
nested within 111 persons.
The final sample comprised 111 persons (46.8% women) from
30 local organizations; the number of participants per organization
ranged between 1 and 17. Participants came from a broad range of
occupational backgrounds, with most participants working as com-
mercial clerks (21.6%), employment center employees (15.3%),
engineers or information technology specialists (14.6%), local
government employees (11.7%), bank employees (12.6%), secre-
taries (8.1%), or social workers (5.4%). Mean age was 39.3 years
(SD !10.6); mean job tenure was 14.7 years (SD !11.0). Most
participants had flexible schedules, with an average contract work
time of 38.2 hr per week (SD !5.1). Average overtime per week
was 4.7 hr (SD !4.5). About one third (33.9%) of the participants
had a leadership position. With respect to family status, 62.6%
lived with a partner, 29.0% lived alone, 3.7% were single parents,
and 4.7% lived with another person who was neither their partner
nor their child. Among all participants, 41.4% had children (M!
1.7, SD !0.7).
Measures
We collected data at the day and the person level. Table 1 shows
the means, standard deviations, and correlations between the study
variables. All items were in German. Participants provided their
responses on 5-point Likert scales; the response format for all
items, except state negative affect and job control, was 1 !I fully
disagree;5!I fully agree.
Day-level measures. We collected day-level data in the
morning (recovery level in the morning) and at the end of the
workday (recovery level at the end of the workday, job demands,
situational constraints, work engagement, and state negative af-
fect).
Recovery level in the morning. We assessed participants’
morning recovery levels with four items from the measure devel-
oped by Sonnentag and Kruel (2006). This measure assesses a
person’s momentary state of being recovered: “This morning, I
feel mentally recovered,” “This morning I feel physically recov-
ered,” “This morning I feel well-rested,” “This morning, I am full
of new energy.” Cronbach’s alpha computed separately for the 4
days of data collection ranged between .89 and .91 (M!.90).
Recovery level at the end of the workday. For measuring
recovery level at the end of the workday, we used the same items
as in the morning, yet this time with reference to the end of the
workday (e.g., “Now, at the end of the workday, I feel mentally
recovered”). Cronbach’s alpha ranged between .83 and .92 (M!
.88).
Job demands. For measuring day-level job demands, we
used three items from a scale developed by (Semmer, 1984; Zapf,
1993) that focused on quantitative demands. We adjusted the items
for day-level measurement. A sample item is “Today I was re-
quired to work fast.” Cronbach’s alpha ranged between .87 and .91
(M!.88).
Situational constraints. We measured day-level situational
constraints with four items from the measure developed by (Sem-
mer, 1984; Zapf, 1993). Again, we rephrased the items slightly so
that they captured the day-specific level of situational constrains.
A sample item is “Today I had to work with materials and
information that were incomplete and outdated.” Cronbach’s alpha
ranged between .83 and .89 (M!.87).
Work engagement. We assessed day-level work engagement
with the nine-item version of the Utrecht Work Engagement Scale
(UWES; Schaufeli, Bakker, & Salanova, 2006) adapted for day-
level assessment. Sample items are “During my work, I felt strong
Table 1
Means, Standard Deviations, and Zero-Order Correlations of All Study Variables
Variable M
a
SD
a
M
b
SD
b
123456789
1. Job control 3.65 0.71 —
2. General level of recovery 2.30 0.75 .18 —
3. General level of work engagement 3.76 0.97 .38 .16 —
4. Recovery level in the morning 3.45 0.66 3.47 0.80 .25 .16 .37 — .28 ".03 ".11 .18 ".19
5. Day-level work engagement 3.23 0.66 3.25 0.70 .31 .15 .55 .43 — ".05 ".24 .46 ".22
6. Day-level job demands 2.67 0.99 2.62 1.12 ".02 .03 ".02 ".15 ".02 — .40 ".18 .22
7. Day-level situational constraints 1.77 0.81 1.79 0.91 ".17 ".06 ".09 ".17 ".24 .43 — ".28 .43
8. Recovery level at the end of workday 2.51 0.72 2.52 0.87 .28 .36 .26 .27 .56 ".21 ".26 — ".25
9. Negative affect at the end of workday 1.30 0.34 1.31 0.42 ".10 ".03 ".05 ".34 ".16 .25 .48 ".18 —
Note. Correlations below the diagonal are person-level correlations (N!111). Correlations above the diagonal are day-level correlations (n!325).
Person-level correlations of r!.24 are significant with p#.01, and those of r!.19 are significant with p#.05. Day-level correlations of r!.11 are
significant with p#.01, and those of r!.13 are significant with p#.05.
a
Means and standard deviations at the person level.
b
Means and standard deviations at the day level.
846 SONNENTAG ET AL.

and vigorous today,” “Today, I was enthusiastic about my job,”
“Today, when I was working, I forgot everything else around me.”
Cronbach’s alpha ranged between .87 and .91 (M!.90).
State negative affect. We used state negative affect at the end
of the workday as a control variable when predicting the recovery
level at the end of the workday. We included this control variable
because the predictor and the outcome variable were assessed at
the same point in time, and we wanted to rule out that significant
associations between these two variables could be attributed to the
specific measurement occasion. Participants responded to
negative-affect items from the Positive and Negative Affect
Schedule (Watson, Clark, & Tellegen, 1988) on a 5-point Likert
scale ranging from 1 !not at all to5!very much. To limit the
time needed to complete the day-level measure, we used six items
(Distressed, Upset, Irritable, Nervous, Jittery, and Afraid) as done
in earlier research (Sonnentag, Binnewies, & Mojza, 2008). Cron-
bach’s alpha ranged between .60 and .79 (M!.69).
Construct validity. To examine whether the five variables
assessed at the end of the workday (i.e., work engagement, recov-
ery at the end of the workday, situational constraints, job demands,
and negative affect) constitute distinct constructs, we conducted a
set of multilevel confirmatory factor analyses using Mplus 6.1. A
five-factor model with all items loading on their respective factors
yielded an acceptable fit ($
2
!546.385; df !289; p#.001;
comparative fit index [CFI] !0.92; root-mean-square error of
approximation [RMSEA] !0.052), and all factor loadings were
significant except for one negative-affect item (Afraid) that had a
marginally significant factor loading (p!.064).
1
All standardized
factor loadings were above .40 except for three items that assessed
state negative affect. This five-factor model fit the data better than
the best fitting four-factor model, with situational constraints and
negative affect loading on one factor ($
2
!719.133; df !293; p#
.001; CFI !0.87; RMSEA !0.069; Satorra–Bentler scaled $
2
[S-B $
2
]!83.8823; df !4; p#.001); the best fitting three-factor
model ($
2
!1123.484; df !296; CFI !0.747; RMSEA !0.093,
S-B $
2
!446.2800; df !7; p#.001); the best fitting two-factor
model ($
2
!1556.545; df !298; CFI !0.399; RMSEA !0.142,
S-B $
2
!621.0137; df !9; p#.001); and a one-factor model
($
2
!2266.581; df !299; CFI !0.399; RMSEA !0.142); S-B
$
2
!1,288.64; df !10; p#.001). Thus, the five end-of-workday
variables clearly represent distinct constructs.
Moreover, to establish that the recovery items (including the
item “I am full of new energy”) did not incidentally measure work
engagement, we conducted an exploratory factor analysis of re-
covery and engagement items using the two-level option in Mplus.
This analysis (conducted separately for recovery data assessed in
the morning and recovery data assessed at the end of the workday)
showed a two-factor solution in which the four recovery items had
an average loading of .83 (morning) and .80 (end of workday) onto
the recovery factor, and a maximum cross-loading onto the work
engagement factor of .12 (morning) and .15 (end of workday).
Person-level measures. At the person level, we assessed a
person’s general level of recovery and general level of work
engagement. For assessing the general level of recovery, we used
the four-item scale developed by Sonnentag and Kruel (2006;
sample item: “During my free time, I feel mentally recovered”).
Cronbach’s alpha was .86. We measured general level of work
engagement with the nine-item version of the UWES (Schaufeli et
al., 2006). Cronbach’s alpha was .91. We included these variables
as control variables in order to take person-level differences into
account when predicting day-specific work engagement and the
day-specific recovery level at the end of the workday. Because
work engagement and level of recovery may depend on job control
experienced (Bakker & Demerouti, 2007; van Veldhoven & Slui-
ter, 2009), we also controlled for job control. We used five items
from the scale developed by (Semmer, 1984; Zapf, 1993), to be
answered on a 5-point Likert scale (1 !very little;5!to a high
degree). A sample item is “Can you influence the way in which
you accomplish your tasks?” Cronbach’s alpha was .78.
Results
Our data set comprised data at the person level (e.g., person-
level control variables) and at the day level (e.g., day-specific work
engagement), with day-level data nested within persons. To take
the noninterdependence of these data into account, we used hier-
archical linear modeling to analyze the data.
Descriptive Analysis
Before testing our hypotheses, we examined within- and
between-person variation in our two outcome variables (work
engagement and recovery level at the end of the workday). As can
be seen in the null model for work engagement (see Table 2),
within-person variance (Level 1 variance) was 0.164 and between-
person variance (Level 2 variance) was 0.345, resulting in a total
variance of 0.509. Thus, for work engagement, within-person
variance was 32.2% and between-person variance was 67.8%. For
recovery level at the end of the workday (see Table 3), within-
person variance was 0.383 and between-person variance was
0.360, corresponding to 51.5% and 48.5%, respectively.
Test of Hypotheses
To test our hypotheses, we compared several nested models.
The null model included only the intercept. In Model 1, we
included control variables; in Model 2, we entered the main
effects; and in Model 3, we included the interaction effects. We
tested the improvement of each model over the previous one by
using the difference between the respective likelihoods. This dif-
ference follows a chi-square distribution, with the degrees of
freedom corresponding to the number of parameters added to the
model.
Hypotheses 1 to 3 proposed that morning recovery level will
predict work engagement and that day-specific job demands and
day-specific situational constraints will moderate the relation be-
tween morning recovery level and work engagement. Table 2
shows the results from hierarchical linear models predicting day-
specific work engagement from morning recovery level, job stres-
sors (job demands, situational constraints), and the interaction
terms between recovery level and job stressors (job demands,
situational constraints). Model 1—including job control and gen-
eral level of work engagement as control variables—showed a
better model fit than the null model. A person’s general level of
work engagement was a highly significant predictor of day-
1
We repeated all analyses, including the hypotheses tests, with a five-
item measure of negative affect. All results remained unchanged.
847
RECOVERY AND WORK ENGAGEMENT

specific work engagement. Model 2—including the main effects—
showed a further improvement over Model 1, with all three pre-
dictor variables being significant. Morning recovery level and job
demands were positively related and situational constraints were
negatively related to work engagement. Model 3—including the
interaction terms—fit the data significantly better than Model 2.
The interaction term between morning recovery level and situa-
tional constraints was significant.
2
To gain more insight into the
pattern of this interaction, we performed simple slope tests
(Preacher, Curran, & Bauer, 2006). As Figure 2 illustrates, on days
with high levels of situational constraints (one SD above the
mean), morning recovery level was not related to work engage-
ment; (%!"0.062; SE !0.079; z!"0.782; ns), but on days
with low levels of situational constraints (one SD below the mean),
morning recovery level was positively related to work engagement
(%!0.184; SE !0.066; z!2.811; p#.01). The interaction term
between morning recovery and job demands, however, was not
significant. Taken together, our data provided support for Hypoth-
eses 1 and 3 but not for Hypothesis 2.
Hypotheses 4 and 5 referred to the prediction of recovery level
at the end of the workday. We hypothesized that work engagement
during the day will be positively related to the recovery level at the
end of the workday (Hypothesis 4) and that situational constraints
will moderate the relation between work engagement and recovery
level at the end of the workday (Hypothesis 5). In these analyses,
we controlled for job control, general level of recovery, morning
recovery level, and negative affect at the end of the workday
because all these variables may have an impact on recovery level
at the end of the workday. When testing the interaction effect
between work engagement and situational constraints, we pursued
a conservative strategy and also included the interaction effect
between work engagement and job demands.
Table 3 shows the results. Model 1—including the control
variables—showed a better model fit than the null model. Job
control and general level of recovery were positively related to
recovery level at the end of the workday. Negative affect at the end
of the workday showed a strong negative association with recovery
level at the end of the workday.
3
When we entered work engage-
ment, job demands, and situational constraints into Model 2, model
fit further improved. Work engagement was positively related to
recovery level at the end of the workday, providing support for
Hypothesis 4. Neither job demands nor situational constraints were
significantly related to recovery level at the end of the workday.
Model 3—including the interaction effects— had a better model fit
than Model 2. The estimate of the interaction between work
engagement and situational constraints was significant.
4
Simple
slope tests (Preacher et al., 2006) showed that on days when
situational constraints were low (one SD below the mean), work
engagement during the workday had a strong positive association
with the recovery level at the end of the workday (%!0.542; SE !
0.125; z!4.335; p#.001). On days with a high level of
situational constraints (one SD above the mean), work engagement
was not related to the recovery level at the end of the workday
(%!0.140; SE !0.142; z!0.992, ns; cf. Figure 3). Overall,
these findings support Hypothesis 5.
Discussion
Our study showed that morning recovery level predicted work
engagement during the workday and that work engagement, in
2
When only the interaction term between morning recovery level and
situational constraints were entered in Model 3, the results remained very
similar and the interaction term was significant (%!"0.383, SE !0.149,
t!"2.570, p#.05).
3
Scholars do not agree whether one should control for negative affect
(Spector, 2006). Therefore, we also ran the models predicting recovery
level at the end of the workday without controlling for state negative affect.
The results remained unchanged. Similarly, when omitting job control
from the control variables, our results did not change. Tables are available
from the first author upon request.
4
When only the interaction term between work engagement and situa-
tional constraints was entered in Model 3, the results did not change (%!
"0.465, SE !0.194, t!"3.212, p#.01, for the interaction term).
Table 2
Multilevel Estimates for Models Predicting Day-Specific Work Engagement
Parameter
Null model Model 1 Model 2 Model 3
Estimate SE t Estimate SE t Estimate SE t Estimate SE t
Intercept 3.241 0.061 52.13 3.246 0.051 62.65 3.240 0.051 63.53 3.237 0.051 63.47
Job control 0.089 0.079 1.13 0.085 0.079 1.08 0.082 0.079 1.04
General level of work engagement 0.343 0.057 6.02
!!
0.343 0.057 6.02
!!
0.338 0.057 5.93
!!
Recovery level in the morning (REC) 0.091 0.042 2.17
!
0.062 0.042 1.48
Job demands (JD) 0.090 0.038 2.37
!
0.107 0.038 2.82
!!
Situational constraints (SC) "0.161 0.063 "2.56
!
"0.155 0.062 "2.50
!
REC &JD "0.089 0.073 "1.22
REC &SC "0.324 0.156 "2.08
!
"2
!
log (lh) 547.899 506.409 492.136 484.170
Diff "2
!
log 41.100
!!
14.273
!!
7.966
!
df 232
Level 1 intercept variance (SE) 0.164 (0.016) 0.164 (0.016) 0.153 (0.015) 0.147 (0.014)
Level 2 intercept variance (SE) 0.345 (0.055) 0.222 (0.038) 0.224 (0.038) 0.229 (0.039)
Note. Job control and general level of work engagement are person-level (Level 2) variables; all other predictor variables are day-level (Level 1) variables.
!
p#.05.
!!
p#.01.
848 SONNENTAG ET AL.

turn, predicted recovery level at the end of the workday. These
reciprocal relations between recovery level and work engagement
did not occur under all circumstances. Situational constraints at-
tenuated the association between morning recovery level and work
engagement during the day and between work engagement during
the day and subsequent recovery level. Although our data cannot
demonstrate causality in a strict sense, the pattern of findings
might imply that recovery level and work engagement mutually
reinforce each other: The more recovered an employee is in the
morning, the more engagement the employee will experience at
work, which limits the decrease in the employee’s recovery level
over the course of the day. Situational constraints interrupt these
reciprocal processes between recovery level and work engage-
ment.
Our findings, focusing on within-person fluctuations of re-
covery level and work engagement, extend results from studies
conducted at the between-person level. These studies have
identified reciprocal associations between resources, such as
optimism and pride in one’s profession on the one hand and
work engagement on the other hand (Hakanen, Perhoniemi, &
Toppinen-Tammer, 2008; Xanthopoulou et al., 2009a). Thus,
when recovery level is conceptualized as a resource (Binnewies
et al., 2009), our findings point in a similar direction: Re-
sources, including recovery level, facilitate work engagement,
Figure 2. Prediction of work engagement. Figure 3. Prediction of recovery level at the end of the workday.
Table 3
Multilevel Estimates for Models Predicting Day-Specific Recovery Level at End of Workday
Parameter
Null model Model 1 Model 2 Model 3
Estimate SE t Estimate SE t Estimate SE t Estimate SE t
Intercept 2.517 0.068 37.01 2.507 0.061 41.10 2.499 0.061 40.97 2.486 0.061 40.75
Job control 0.196 0.091 2.15
!
0.185 0.090 2.06
!
0.189 0.090 2.10
!
General level of recovery 0.307 0.086 3.57
!!
0.308 0.085 3.62
!!
0.301 0.084 3.58
!!
Recovery level in the morning 0.145 0.062 2.34
!
0.109 0.061 1.79 0.095 0.061 1.56
Negative affect at end of workday "0.565 0.130 "4.35
!!
"0.403 0.135 "2.99
!!
"0.424 0.135 "3.14
!!
Work engagement (WE) 0.359 0.102 3.52
!!
0.341 0.103 3.31
!!
Job demands (JD) "0.064 0.057 "1.12 "0.050 0.057 "0.88
Situational constraints (SC) "0.029 0.095 "0.30 "0.100 0.098 "1.02
WE &JD 0.089 0.155 0.57
WE &SC "0.525 0.222 "2.39
!
"2
!
log (lh) 752.949 710.977 697.919 691.891
Diff "2
!
log 41.972
!!
13.058
!!
6.028
!
df 432
Level 1 intercept variance (SE) 0.383 (0.037) 0.347 (0.033) 0.329 (0.032) 0.323 (0.031)
Level 2 intercept variance (SE) 0.360 (0.069) 0.287 (0.057) 0.287 (0.056) 0.282 (0.055)
Note. Job control and general level of recovery are person-level (Level 2) variables; all other predictor variables are day-level (Level 1) variables.
!
p#.05.
!!
p#.01.
849
RECOVERY AND WORK ENGAGEMENT

which in turn helps to keep resources at a higher level than
when work engagement is low. We have to note that in absolute
terms, recovery level goes down during the course of the
workday. However, one might speculate that over a series of
days with high levels of morning recovery and high levels of
work engagement during the day, gain cycles might occur that
are reflected in increasing recovery levels and increasing levels
of work engagement over the course of several days.
We identified situational constraints as harmful moderators in
the recovery– engagement process. On days when situational con-
straints were high, the reciprocal positive associations between
recovery level and work engagement broke down. One can think of
at least two underlying mechanisms that render situational con-
straints detrimental: The first mechanism refers to the affective
consequences of situational constraints. Situational constraints
evoke negative affective states, such as anger and anxiety (Rodell
& Judge, 2009), that call for emotion regulation. Emotion regula-
tion may interfere with work engagement and will consume addi-
tional resources so that the recovery level will drop. This mecha-
nism should also apply to job demands; however, no moderator
effects were found for job demands.
5
Thus, a second mechanism
seems more likely: Situational constraints, such as lack of infor-
mation or supplies, actually impede the task completion process,
requiring additional effort to get the work done. As a consequence,
when one encounters situational constraints, it is more difficult to
become or stay engaged (i.e., energetic, absorbed) and to keep
one’s recovery level.
Importantly, job demands did not turn out as a moderator,
neither in the association between morning recovery level and
work engagement nor in that between work engagement and
recovery level at the end of the workday. Although we did not
propose an interaction effect for the latter association (i.e., between
work engagement and recovery level at the end of the workday),
not finding an interaction effect for the first association (between
morning recovery level and work engagement) was unexpected. It
has to be noted that job demands did have a main positive effect on
work engagement, implying that job demands and recovery level
had an additive, but not a multiplicative, effect on work engage-
ment.
One reason why the interaction between morning recovery level
and job demands was not significant may be the fact that the
absolute level of job demands was not very high in the present
sample (M!2.67 on a 5-point scale). Even on days when
employees did not feel so recovered, an increase in job demands
might still have fueled their engagement. It is conceivable that the
interaction would have been found if the average level of job
demands was higher, because on very demanding days (as com-
pared to not-so-demanding days) employees would profit more
from their morning recovery level.
We found a negative relation between situational constraints and
work engagement and a positive relation between job demands and
work engagement. These within-person results reflect findings
from a meta-analysis by Crawford et al. (2010) that used between-
person correlations to analyze the association between various
types of job stressors and work engagement. Our job demand
measure largely overlaps with Crawford et al.’s notion of chal-
lenge demands (e.g., high workload), and our situational constraint
measure captures core aspects of hindrance demands. In combi-
nation with the results of our moderator analyses, the overall
pattern of findings suggests that these types of job stressors have
rather distinct implications for work engagement. Whereas chal-
lenging demands tend to stimulate work engagement, hindrances
and constraints make work engagement rather unlikely.
Recently, researchers initiated a debate on whether there is a
dark side to being too engaged (Bakker, Albrecht, & Leiter, 2011;
Halbesleben et al., 2009). Hablesleben et al. found that employees
with high levels of work engagement reported that work was
interfering with their family life. Our study did not directly assess
work–family conflict or a related construct. Our data, however,
showed that at least at the day level, work engagement did not put
employees in a poor affective or energetic state that might interfere
with their family or other nonwork responsibilities. Rather, our
data suggest that after a highly engaged day at work, employees
leave the workplace in a more recovered state than that experi-
enced on a day with less work engagement. This state should
enable them to engage in nonwork activities (cf. Rothbard, 2001),
again relatively more than on days with low levels of work
engagement. In a broader context, our findings are compatible with
a work–family enrichment perspective (Greenhaus & Powell,
2006). Experiencing a positive state at work (e.g., work engage-
ment) fosters positive states at the end of the workday that, in turn,
will have a positive impact on nonwork life.
Limitations and Directions for Future Research
Our study is subject to several limitations that suggest directions
for future research. First, we assessed our data with self-report
measures, which might lead to concerns about common method
variance (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). How-
ever, because we focused on within-person fluctuations, results
cannot be explained by individual differences. Moreover, for some
part of the analyses we assessed predictor and outcome variables at
different points in time, further reducing the likelihood that our
findings are due to common method bias. Because we assessed
work engagement as a predictor and recovery at the end of the
workday at the same point in time, we controlled for concurrent
negative affect in order to rule out that the momentary affect drove
the significant relation between work engagement and recovery
level. To overcome problems potentially associated with self-
report measures, future studies might assess job stressors by co-
worker reports or use more objective indicators of job demands
and constraints (e.g., number of interruptions by phone calls).
Second, although we had two measurement occasions per day,
our correlational design does not allow any conclusions about
causality in a strict sense. For instance, the association between
work engagement and recovery level at the end of the workday
might reflect a common cause, such as positive events at work.
Although we cannot rule out this possibility completely, our anal-
ysis demonstrated that work engagement predicted end-of-
workday recovery level even when morning recovery level, job
5
Following the suggestion of an anonymous review, we tested whether
negative affect mediated (a) the interaction effect between morning recov-
ery level and situational constraints on work engagement and (b) the
interaction effect between work engagement and situational constraints on
recovery level at the end of the workday. In both analyses, we found no
evidence of mediated moderation. Thus, our empirical data did not support
the negative-affect mechanism.
850 SONNENTAG ET AL.

stressors, and state negative affect were included in the regression
equation. Future studies might control for positive events encoun-
tered during the day and realize a third measurement occasion per
day to further separate the measurement points.
Third, we limited ourselves to two types of job stressors, and
within these types, we used a rather broad conceptualization of
situational constraints. Future studies could assess other types of
job stressors. With respect to situational constraints, a recent study
(Liu, Nauta, Li, & Fan, 2010) suggested differentiating between
job context constraints, such as lack of equipment or supplies, and
interpersonal constraints, such as interruptions by other people or
inadequate help from others. It might be that the moderator effect
can be found only for some types of constraints and not for others.
Fourth, we adopted a day-level approach. Although this ap-
proach adds an important perspective to research on work engage-
ment and offers valuable insights, we do not know how our
day-level findings generalize to other time frames. Therefore,
future research should examine the dynamics between recovery
level and work engagement within longer as well as shorter time
frames. Longer time frames would imply week-level studies or
longer term longitudinal designs, for instance, over several
months. However, narrowing the time frame by using shorter
assessment intervals might increase our understanding of the dy-
namics between work engagement and recovery. For instance, one
might examine how episodes of high work engagement result in
affective and energetic states during a work break and how work
breaks, in turn, help one to become engaged when back at work
again.
Practical Implications and Conclusions
Our study suggests practical implications that may help employ-
ees to be engaged at work. First, it is important that employees feel
recovered when they come to work. Several avenues to a high
recovery level should be considered, including beneficial recovery
experiences at home, such as mental detachment from work and
good sleep (Sonnentag et al., 2008). Managers should try not to
interfere with these experiences, for instance, by respecting bound-
aries between employees’ work and nonwork lives and by avoid-
ing unfair treatment that has been shown to be associated with
insomnia problems (Boswell & Olson-Buchanon, 2007; Green-
berg, 2006).
Furthermore, because situational constraints turned out to be a
factor that reduced the benefits of a high recovery level on work
engagement and of high work engagement on the subsequent
recovery level, it is important to reduce situational constraints at
work, for instance, by job design efforts (Semmer, 2006). How-
ever, as the reduction in situational constraints may not always be
feasible (Holman, Axtell, Sprigg, Totterdell, & Wall, 2010), it is
important to prevent a dramatic drop in employees’ recovery level
on days when situation constraints occur, for example, by sched-
uling breaks that provide sufficient revitalization and by encour-
aging employees to actually take these breaks.
Taken together, our study shows that recovery level and work
engagement are mutually related and probably reinforce each
other. Importantly, this association becomes evident only during
days without high levels of situational constraints, making reduc-
tion of and effective coping with this type of job stressor a crucial
goal within stress management efforts.
References
Bakker, A. B., Albrecht, S., & Leiter, M. P. (2011). Key questions
regarding work engagement. European Journal of Work and Organiza-
tional Psychology, 20, 4 –28. doi:10.1080/1359432X.2010.485352
Bakker, A. B., & Bal, P. M. (2010). Weekly work engagement and
performance: A study among starting teachers. Journal of Occupational
and Organizational Psychology, 83, 189 –206. doi:10.1348/
096317909X402596
Bakker, A. B., & Demerouti, E. (2007). The job demands–resources model:
State of the art. Journal of Managerial Psychology, 22, 309 –328.
doi:10.1108/02683940710733115
Bakker, A. B., Schaufeli, W. B., Leiter, M. P., & Taris, T. W. (2008). Work
engagement: An emerging concept in occupational health psychology.
Work & Stress, 22, 187–200. doi:10.1080/02678370802393649
Bakker, A. B., van Emmerik, I. H., Geurts, S. A. E., & Demerouti, E.
(2008). Recovery turns job demands into challenges: A diary study on
work engagement and performance. Working paper, Erasmus University
Rotterdam, Rotterdam, the Netherlands.
Beal, D. J., Weiss, H. M., Barros, E., & MacDermid, S. M. (2005). An
episodic process model of affective influences on performance. Journal
of Applied Psychology, 90, 1054 –1068. doi:10.1037/0021-
9010.90.6.1054
Binnewies, C., Sonnentag, S., & Mojza, E. J. (2009). Daily performance at
work: Feeling recovered in the morning as a predictor of day-level job
performance. Journal of Organizational Behavior, 30, 67–93. doi:
10.1002/job.541
Bledow, R., Schmitt, A., Frese, M., & Ku¨hnel, J. (2011). The affective shift
model of work engagement. Journal of Applied Psychology, 96, 1246 –
1257. doi:10.1037/a0024532
Boswell, W. R., & Olson-Buchanon, J. B. (2007). The use of communi-
cations technologies after hours: The role of work attitudes and work–
life conflict. Journal of Management, 33, 592– 610. doi:10.1177/
0149206307302552
Crawford, E. R., LePine, J. A., & Rich, B. L. (2010). Linking job demands
and resources to employee engagement and burnout: A theoretical
extension and meta-analytic test. Journal of Applied Psychology, 95,
834 – 848. doi:10.1037/a0019364
de Lange, A. H., Taris, T. W., Kompier, M. A. J., Houtman, I. L. D., &
Bongers, P. M. (2003). “The very best of the millennium”: Longitudinal
research and the demand-control-(support) model. Journal of Occupa-
tional Health Psychology, 8, 282–305. doi:10.1037/1076-8998.8.4.282
Demerouti, E., Bakker, A. B., Nachreiner, F., & Schaufeli, W. B. (2001).
Job demands–resources model of burnout. Journal of Applied Psychol-
ogy, 86, 499 –512. doi:10.1037/0021-9010.86.3.499
Demerouti, E., Bakker, A. B., Sonnentag, S., & Fullagar, C. (2012).
Work-related flow and energy at work and at home: A study on the role
of daily recovery. Journal of Organizational Behavior, 33, 276 –295.
doi: 10.1002/job.760
Gilboa, S., Shirom, A., Fried, Y., & Cooper, C. (2008). A meta-analysis of
work demand stressors and job performance: Examining main and
moderating effects. Personnel Psychology, 61, 227–271. doi:10.1111/
j.1744-6570.2008.00113.x
Greenberg, J. (2006). Losing sleep over organizational injustice: Attenu-
ating insomniac reactions to underpayment inequity with supervisory
training in interactional justice. Journal of Applied Psychology, 91,
58 – 69. doi:10.1037/0021-9010.91.1.58
Greenhaus, J. H., & Powell, G. N. (2006). When work and family are
allies: A theory of work–family enrichment. Academy of Management
Journal, 31, 72–92. doi:10.5465/AMR.2006.19379625
Hakanen, J. J., Perhoniemi, R., & Toppinen-Tammer, S. (2008). Positive
gain spirals at work: From job resources to work engagement, personal
initiative and work-unit innovativeness. Journal of Vocational Behavior,
73, 78 –91. doi:10.1016/j.jvb.2008.01.003
Halbesleben, J. R. B., Harvey, J., & Bolino, M. C. (2009). Too engaged?
851
RECOVERY AND WORK ENGAGEMENT

A conservation of resources view of the relationship between work
engagement and work interference with family. Journal of Applied
Psychology, 94, 1452–1465. doi:10.1037/a0017595
Hockey, G. R. J. (1997). Compensatory control in the regulation of human
performance under stress and high workload: A cognitive-energetical
framework. Biological Psychology, 45, 73–93. doi:10.1016/S0301-
0511(96)05223-4
Holman, D., Axtell, C. M., Sprigg, C. A., Totterdell, P., & Wall, T. D.
(2010). The mediating role of job characteristics in job redesign inter-
ventions: A serendipitous quasi-experiment. Journal of Organizational
Behavior, 31, 84 –105. doi:10.1002/job.631
Ilies, R., Dimotakis, N., & De Pater, I. E. (2010). Psychological and
physiological reactions to high workloads: Implications for well-being.
Personnel Psychology, 63, 407– 436. doi:10.1111/j.1744-6570.2010
.01175.x
Ku¨hnel, J., Sonnentag, S., & Bledow, R. (2012). Resources and time
pressure as day-level antecedents of work engagement. Journal of Oc-
cupational and Organizational Psychology, 85, 181–198. doi:10.111/
j.2044-8325.2011.02022.x
Leitner, K., & Resch, M. G. (2005). Do the effects of job stressors on
health persist over time? A longitudinal study with observational stressor
measures. Journal of Occupational Health Psychology, 10, 18 –30. doi:
10.1037/1076-8998.10.1.18
LePine, J. A., Podsakoff, N. P., & LePine, M. A. (2005). A meta-analytic
test of the challenge stressor– hindrance stressor framework: An expla-
nation for inconsistent relationships among stressors and performance.
Academy of Management Journal, 48, 764 –775. doi:10.5465/
AMJ.2005.18803921
Liu, C., Nauta, M. M., Li, C., & Fan, J. (2010). Comparisons of organi-
zational constraints and their relations to strains in China and the United
States. Journal of Occupational Health Psychology, 15, 452– 467. doi:
10.1037/a0020721
Mauno, S., Kinnunen, U., & Ruokolainen, M. (2007). Job demands and
resources as antecedents of work engagement: A longitudinal study.
Journal of Vocational Behavior, 70, 149 –171. doi:10.1016/
j.jvb.2006.09.002
Newman, D. A., Joseph, D. L., & Hulin, C. L. (2010). Job attitudes and
employee engagement: Considering the attitude “A-factor.” In S. Al-
brecht (Ed.), The handbook of employee engagement: Perspectives,
issues, and research practice (pp. 43– 61). Cheltenham, England: Elgar.
Peters, L. H., & O’Connor, E. J. (1980). Situational constraints and work
outcomes: The influences of a frequently overlooked construct. Academy
of Management Review, 5, 391–397.
Peters, L. H., O’Connor, E. J., Eulberg, J. R., & Watson, T. W. (1988). An
examination of situational constraints in Air Force work settings. Human
Performance, 1, 133–144. doi:10.1207/s15327043hup0102_4
Podsakoff, N. P., LePine, J. A., & LePine, M. A. (2007). Differential
challenge stressor– hindrance stressor relationships with job attitudes,
turnover intention, turnover, and withdrawal behavior: A meta-analysis.
Journal of Applied Psychology, 92, 438 – 454. doi:10.1037/0021-
9010.92.2.438
Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, N. P. (2003).
Common method biases in behavioral research: A critical review of the
literature and recommended remedies. Journal of Applied Psychology,
88, 879 –903. doi:10.1037/0021-9010.88.5.879
Preacher, K. J., Curran, P. J., & Bauer, D. J. (2006). Computational tolls for
probing interactions in multiple linear regression, multilevel modeling,
and latent curve analysis. Journal of Educational and Behavioral Sta-
tistics, 31, 437– 448. doi:10.3102/10769986031004437
Rodell, J. B., & Judge, T. A. (2009). Can “good” stressors spark “bad”
behaviors? The mediating role of emotions in links of challenge and
hindrance stressors with citizenship and counterproductive behaviors.
Journal of Applied Psychology, 94, 1438 –1451. doi:10.1037/a0016752
Rothbard, N. P. (2001). Enriching or depleting? The dynamics of engage-
ment in work and family roles. Administrative Science Quarterly, 46,
655– 684. doi:10.2307/3094827
Ryan, R. M., & Deci, E. L. (2000). Self-determination theory and the
facilitation of intrinsic motivation, social development, and well-being.
American Psychologist, 55, 68 –78. doi:10.1037/0003-066X.55.1.68
Salanova, M., Agut, S., & Peiro´, J. M. (2005). Linking organizational
resources and work engagement to employee performance and customer
loyalty: The mediation of service climate. Journal of Applied Psychol-
ogy, 90, 1217–1227. doi:10.1037/0021-9010.90.6.1217
Salanova, M., Llorens, S., & Schaufeli, W. B. (2011). “Yes, I can, I feel
good, and I just do it!” On gain cycles and spirals of efficacy beliefs,
affect, and engagement. Applied Psychology, 60, 255–285. doi:10.1111/
j.1464-0597.2010.00435.x
Salanova, M., Schaufeli, W. B., Xanthopoulou, D., & Bakker, A. B.
(2010). The gain spiral of resources and work engagement: Sustaining a
positive worklife. In A. B. Bakker & M. P. Leiter (Eds.), Work engage-
ment: Recent developments in theory and research (pp. 118 –131). New
York, NY: Psychology Press.
Sanz-Vergel, A. I., Demerouti, E., Moreno-Jime´nez, B., & Mayo, M.
(2010). Work–family balance and energy: A day-level study on recovery
conditions. Journal of Vocational Behavior, 76, 118 –130. doi:10.1016/
j.jvb.2009.07.001
Schaufeli, W. B., & Bakker, A. B. (2004). Job demands, job resources, and
their relationship with burnout and engagement: A multi-sample study.
Journal of Organizational Behavior, 25, 293–315. doi:10.1002/job.248
Schaufeli, W. B., Bakker, A. B., & Salanova, M. (2006). The measurement
of work engagement with a short questionnaire: A cross-national study.
Educational and Psychological Measurement, 66, 701–716. doi:
10.1177/0013164405282471
Schaufeli, W. B., Bakker, A. B., & van Rhenen, W. (2009). How changes
in job demands and resources predict burnout, work engagement and
sickness absenteeism. Journal of Organizational Behavior, 30, 893–917.
doi:10.1002/job.595
Schaufeli, W. B., Salanova, M., Gonza´les-Roma´, V., & Bakker, A. B.
(2002). The measurement of engagement and burnout: A two sample
confirmatory factor analytic approach. Journal of Happiness Studies, 3,
71–92. doi:10.1023/A:1015630930326
Schaufeli, W. B., Taris, T. W., & van Rhenen, W. (2008). Workaholism,
burnout, and work engagement: Three of a kind or three different kinds
of employee well-being? Applied Psychology, 57, 173–203. doi:
10.1111/j.1464-0597.2007.00285.x
Semmer, N. (1984). Streßbezogene Ta¨tigkeitsanalyse [Stress-oriented task-
analysis]. Weinheim, Germany: Beltz.
Semmer, N. K. (2006). Job stress interventions and the organization of
work. Scandinavian Journal of Work Environment and Health, 32,
515–527. doi:10.5271/sjweh.1056
Shirom, A. (2011). Vigor as a positive affect at work: Conceptualizing
vigor, its relations with related constructs, and its antecedents and
consequences. Review of General Psychology, 15, 50 – 64. doi:10.1037/
a0021853
Simbula, S. (2010). Daily fluctuations in teachers’ well-being: A diary
study using the job demands–resources model. Anxiety, Stress, & Cop-
ing, 23, 563–584. doi:10.1080/10615801003728273
Sonnentag, S. (2003). Recovery, work engagement, and proactive behav-
ior: A new look at the interface between non-work and work. Journal of
Applied Psychology, 88, 518 –528. doi:10.1037/0021-9010.88.3.518
Sonnentag, S., Binnewies, C., & Mojza, E. J. (2008). “Did you have a nice
evening?” A day-level study on recovery experiences, sleep, and affect.
Journal of Applied Psychology, 93, 674 – 684. doi:10.1037/0021-
9010.93.3.674
Sonnentag, S., Dormann, C., & Demerouti, E. (2010). Not all days are
created equal: The concept of state work engagement. In A. B. Bakker
& M. P. Leiter (Eds.), Work engagement: Recent developments in theory
and research (pp. 25–38). New York, NY: Psychology Press.
852 SONNENTAG ET AL.

Sonnentag, S., & Jelden, S. (2009). Job stressors and the pursuit of sport
activities: A day-level perspective. Journal of Occupational Health
Psychology, 14, 165–181. doi:10.1037/a0014953
Sonnentag, S., & Kruel, U. (2006). Psychological detachment from work
during off-job time: The role of job stressors, job involvement, and
recovery-related self-efficacy. European Journal of Work and Organi-
zational Psychology, 15, 197–217. doi:10.1080/13594320500513939
Spector, P. E. (2006). Method variance in organizational research: Truth or
urban legend? Organizational Research Methods, 9, 221–232. doi:
10.1177/1094428105284955
Spector, P. E., & Jex, S. M. (1998). Development of four self-report
measures of job stressors and strain: Interpersonal Conflict at Work
Scale, Organizational Constraints Scale, Quantitative Workload Inven-
tory, and Physical Symptoms Inventory. Journal of Occupational Health
Psychology, 3, 356 –367. doi:10.1037/1076-8998.3.4.356
Trougakos, J. P., & Hideg, I. (2009). Momentary work recovery: The role
of within-day work breaks. In S. Sonnentag, P. L. Perrewe´, & D. C.
Ganster (Eds.), Current perspectives on job-stress recovery (pp. 37– 84).
Bingley, England: Emerald.
van Beek, I., Taris, T. W., & Schaufeli, W. B. (2011). Workaholic and
work engagement employees: Dead ringers or worlds apart? Journal of
Occupational Health Psychology, 16, 468 – 482. doi:10.1037/a0024392
van Veldhoven, M. J. P. M., & Sluiter, J. K. (2009). Work-related recovery
opportunities: Testing scale properties and validity in relation to health.
International Archives of Occupational and Environmental Health, 82,
1065–1075. doi:10.1007/s00420-009-0411-z
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and vali-
dation of brief measures of positive and negative affect: The PANAS
scales. Journal of Personality and Social Psychology, 54, 1063–1070.
doi:10.1037/0022-3514.54.6.1063
Xanthopoulou, D., Bakker, A. B., Demerouti, E., & Schaufeli, W. B.
(2009a). Reciprocal relationships between job resources, personal re-
sources, and work engagement. Journal of Vocational Behavior, 74,
235–244. doi:10.1016/j.jvb.2008.11.003
Xanthopoulou, D., Bakker, A. B., Demerouti, E., & Schaufeli, W. B.
(2009b). Work engagement and financial returns: A diary study on the
role of job and personal resources. Journal of Occupational and Orga-
nizational Psychology, 82, 183–200. doi:10.1348/096317908X285633
Xanthopoulou, D., Bakker, A. B., Heuven, E., Demerouti, E., & Schaufeli,
W. B. (2008). Working in the sky: A diary study on work engagement
among flight attendants. Journal of Occupational Health Psychology,
13, 345–356. doi:10.1037/1076-8998.13.4.345
Zapf, D. (1993). Stress-oriented analysis of computerized office work.
European Work and Organizational Psychologist, 3, 85–100.
Zijlstra, F. R. H., & Cropley, M. (2006). Recovery after work. In F. A.
Jones, R. J. Burke, & M. Westman (Eds.), Work–life balance: A psy-
chological perspective (pp. 219 –234). New York, NY: Psychology
Press.
Zohar, D., Tzischinski, O., & Epstein, R. (2003). Effects of energy avail-
ability on immediate and delayed emotional reactions to work events.
Journal of Applied Psychology, 88, 1082–1093. doi:10.1037/0021-
9010.88.6.1082
Received February 22, 2011
Revision received January 24, 2012
Accepted January 26, 2012 !
E-Mail Notification of Your Latest Issue Online!
Would you like to know when the next issue of your favorite APA journal will be available
online? This service is now available to you. Sign up at http://notify.apa.org/ and you will be
notified by e-mail when issues of interest to you become available!
853
RECOVERY AND WORK ENGAGEMENT